Automatically generated learned_params section from scenario params file in default config
When generating default config files for preprocessing/training configs, it would be useful to use the names of actual parameters mentioned in the scenario params CSV file to generate the example learned_params
section (currently it just uses some dummy parameters named param1
and param2
).
By default most parameters would be classified as "type: regression", but we might also be able to use some simple heuristics for guessing "type: classification" (e.g. if the parameter values are are all drawn from a set of integers).
It could instead list all params mentioned in the CSV file. This might be useful to also combine with #70, especially if there are a large number of params.
This also relates directly to #79.